AI Algorithms and Foundation Models
We explore and apply cutting-edge artificial intelligence algorithms and trustworthy AI platform technologies. Our mission is to build foundational models and platforms – including Large Language Models (LLMs), Multimodal Generative AI, and Generative Agents – and to securely integrate distributed, large-scale biomedical data from multiple sources. This work establishes a trustworthy AI infrastructure for smart healthcare, fostering a next-generation ecosystem of foundational models and intelligent science. Our current key focus areas include foundational models, deep learning, reinforcement learning, Natural Language Processing (NLP), Computer Vision (CV), Knowledge Graphs (KG), search, recommendation systems, federated learning, large-scale deep learning cloud-native platforms, and a trustworthy AI operating system.
End-to-End Intelligent Drug Discovery
Leveraging cutting-edge artificial intelligence technologies—particularly deep learning, reinforcement learning, generative AI, and foundation models—we conduct innovative drug discovery research. This research encompasses target identification, structure design, intelligent generation, high-throughput and virtual screening, efficacy and toxicity analysis, patient candidate selection, and prognosis prediction. By integrating findings from animal and clinical studies, we establish an intelligent analysis and decision-making feedback loop that bridges dry and wet lab experiments. We are developing a large-scale virtual drug screening platform and a full-process, self-feedback, closed-loop intelligent system to advance the digitalization of drug discovery. This effort aims to accelerate the development of domestically produced innovative drugs for high-challenge diseases, such as highly lethal and rare cancers. Our current focus lies in designing small molecule drugs, functional nucleic acids (including aptamers, nucleic acid vaccines, and drugs), and protein antibodies.
Multimodal Intelligent Diagnosis and Treatment
We conduct research and development on intelligent clinical decision-support algorithms and platforms encompassing the entire lifecycle of smart medical practices. By leveraging the latest AI technologies—including deep learning, reinforcement learning, natural language processing (NLP), computer vision (CV), and knowledge graphs—we investigate core technologies for multimodal data fusion. Our work aims to enhance the accuracy of multi-dimensional data reconstruction and build integrated analysis models that synthesize imaging, pathology, clinical data, and biomolecular information (such as genomics, proteomics, and single-cell data). Furthermore, we are developing a foundational model for medicine to comprehensively support clinical intelligence. This initiative involves creating algorithms and platforms for intelligent diagnosis, proteomics, multimodal and multi-omics analysis, AI-powered early cancer screening, liquid biopsy, digital pathology, drug clinical prognosis, companion diagnostics, and clinical decision support.
Prof. Xiaolin Andy Li
Chief AI Scientist, Hangzhou Institute of Medicine, Chinese Academy of Sciences.
Previously, he served as a Tenured Full Professor and Director of the Computer Engineering Division at the University of Florida, USA. He spearheaded the creation of the U.S. National Deep Learning Center (a consortium of four universities: UF, CMU, UO, UMKC, involving over 80 internationally renowned professors and 30+ corporate members including NVIDIA, Google, Eli Lilly, Baidu, Tencent, NetEase, and Ping An), serving as its Founding Director. In 2010, he received the prestigious NSF CAREER Award. He was honored as one of China's "Top Ten AI Figures of the Year" in 2021, and has received multiple best paper awards at leading conferences/journals, as well as championships in international challenges. He led the development of the United States' fastest 200G campus research network (2012) and its first software-defined intelligent campus cloud platform. He played a key role in facilitating a multi-ten-million-dollar donation from NVIDIA, contributing to the establishment of the world's top-ranked university AI computing center (2020). He also led the successful application for the Ministry of Science and Technology's National New Generation AI Open Innovation Platform (2022). He has led the development of numerous influential AI algorithms and platforms, including:Intelligent Platforms: CognitiveEngine, DeepCloud, GatorCloud ,Protein Folding & Target Discovery: PrimateAI, DeepFolding, FoldingZero ,Intelligent Diagnostics: DeepBipolar, DeepCancer, MySurgeryRisk, Dr. Copilot ,Intelligent Drug Discovery: DeepAtom, DyScore, DrugMetric, BatmanNet, AptaDiff, RNADiffFold, AtomicFold, AtomicVac ,Research Assistants: Insage, ScholarClub ,Federated Learning: FLEX, iBond He has authored more than 200 peer-reviewed publications in top-tier international journals and conferences spanning deep learning, biomedicine, cloud computing, and security/privacy. His research outcomes have been successfully applied to target discovery, protein and nucleic acid structure prediction, molecular generation, genomics, proteomics, pathological imaging, and clinical decision support systems. Notably, many of his former students have become tenured professors at prestigious universities in the United States, including a top-five ranked pharmacy school.
